<?xml version="1.0"?>
<tool id="analysis_load_gff3" profile="16.04" name="Load a GFF3 annotation file" version="@WRAPPER_VERSION@.0">
    <description>into Tripal</description>
    <macros>
        <import>macros.xml</import>
    </macros>
    <expand macro="requirements"/>
    <code file="tripal.py"/>
    <expand macro="stdio"/>
    <command><![CDATA[
        tmp_dir=`mktemp -d -p "@DATA_DIR@"` && chmod a+rx "\${tmp_dir}"

        &&

        cp '${gff3}' "\${tmp_dir}/annotation.gff3"

        &&

        @AUTH@

        tripaille analysis load_gff3

            --organism_id '${organism}'

            --analysis_id '${analysis}'

            --import_mode '${import_mode}'

            #if str($naming.mode) == 'manual':
                --re_mrna '${naming.re_mrna}'
                --re_protein '${naming.re_protein}'
            #end if

            #if str($advanced.landmark_type):
                --landmark_type '${advanced.landmark_type}'
            #end if

            #if str($advanced.alt_id_attr):
                --alt_id_attr '${advanced.alt_id_attr}'
            #end if

            ${advanced.create_organism}

            #if str($target.target_organism):
                --target_organism '${target.target_organism}'
            #end if

            #if str($target.target_type):
                --target_type '${target.target_type}'
            #end if

            ${target.target_create}

            "\${tmp_dir}/annotation.gff3"

        &&

        echo "Data loaded" > '$results'
    ]]></command>
    <inputs>
        <param name="gff3"
               type="data"
               format="gff3"
               label="Annotation file" />

        <param argument="--organism"
               type="select"
               dynamic_options="list_organisms()"
               label="Organism" />

        <param argument="--analysis"
               type="select"
               dynamic_options="list_analyses()"
               label="Analysis" />

        <param name="import_mode"
               argument="--import_mode"
               type="select"
               label="Loading method">
            <option value="update" selected="true">Import everything, update when already existing</option>
            <option value="add_only">Import only new features</option>
        </param>

        <conditional name="naming">
            <param name="mode"
                   type="select"
                   label="Naming method for proteins"
                   help="in case your GFF file does not contain polypeptide features">
                <option value="auto">Automatic</option>
                <option value="manual">Manual</option>
            </param>
            <when value="auto"/>
            <when value="manual">
                <param name="re_mrna"
                       argument="--re_mrna"
                       type="text"
                       label="Regular expression for the mRNA name"
                       help="this regex will be used to extract parts of the mRNA feature name">
                    <expand macro="sanitized"/>
                </param>

                <param name="re_protein"
                       argument="--re_protein"
                       type="text"
                       label="Replacement string for the protein name"
                       help="will be used to generate the protein name based on the mRNA name">
                    <expand macro="sanitized"/>
                </param>
            </when>
        </conditional>

        <section name="advanced" title="Advanced options" expanded="False">
            <param name="landmark_type"
                   argument="--landmark_type"
                   type="text"
                   optional="true"
                   label="Landmark type"
                   help="A Sequence Ontology type for the landmark sequences in the GFF fie (e.g. \'chromosome\'). Will be used to create them if they don't already exist." />

            <param name="alt_id_attr"
                   argument="--alt_id_attr"
                   type="text"
                   optional="true"
                   label="ID attribute"
                   help="Name of the GFF attribute that contains a unique identifier for each feature. Leave empty to use the 'ID' attribute" />

            <param name="create_organism"
                   argument="--create_organism"
                   type="boolean"
                   checked="false"
                   truevalue="--create_organism"
                   falsevalue=""
                   label="Create organisms specified in 'organism' attribute"
                   help="If not found, create features referenced in the target attribute." />
        </section>

        <section name="target" title="Target attribute handling" expanded="False">
            <param name="target_organism"
                   argument="--target_organism"
                   type="text"
                   optional="true"
                   label="Target organism name"
                   help="Name of organism corresponding to target attribute. Abbreviation or common name as created with the 'Create Organism' tool." />
            <param name="target_type"
                   argument="--target_type"
                   type="text"
                   optional="true"
                   label="Target feature type"
                   help="Type of features referenced in the target attribute. Should be a Sequence Ontology term." />
            <param name="target_create"
                   argument="--target_create"
                   type="boolean"
                   checked="false"
                   truevalue="--target_create"
                   falsevalue=""
                   label="Create target features"
                   help="If not found, create features referenced in the target attribute." />
        </section>
        <expand macro="wait_for"/>
    </inputs>
    <outputs>
        <data format="txt" name="results" label="Load GFF3 into Tripal" />
    </outputs>
    <tests>
        <test expect_failure="true" expect_exit_code="1">
            <param name="gff3" value="sample.gff3" />
            <param name="organism" value="Testus testus" />
            <param name="analysis" value="Annotation xx" />
            <conditional name="naming">
                <param name="mode" value="manual" />
                <param name="re_mrna" value="([a-z]{4}[0-9]+)_rna" />
                <param name="re_protein" value="([a-z]{4}[0-9]+)_prot" />
            </conditional>

            <expand macro="test_result" />
        </test>
    </tests>
    <help><![CDATA[
        @HELP_OVERVIEW@

        **Load GFF3**

        With this tool, you can load features from a GFF3 file into the Tripal/Chado database.

        @HELP@
    ]]></help>
    <expand macro="citation"/>
</tool>